Focus bracketing and stitching?

Next plans in the world of gigapixels

I’m at a bit of a loss at the moment, so I thought I’d post my thoughts here for others to possibly help answer.

In the aftermath of the Prague 18 Gigapixel madness, I’ve got lots of questions to answer from many people. I promise that I’ll make another post about all the technical aspects of that image – soon!

Also, congratulations are in order to the folks over in Dresden who made the 26 gigapixel image. Great work!

Anyway, what comes next of course is…. a larger image, and a new record 🙂 I can’t help but try it. I won’t make any promises or guarantees though. Since even this Prague gigapixel image was only finished due to a great amount of good fortune, I’ll keep my expectations low on surpassing it in the near future.

Subject matter?

But… ‘What to shoot’ ? It’s easy to make a boring photo of course. The trick in setting a new record for some kind of image is of course finding out what to shoot – and here is the subject of this post. Not everything is easy to shoot when you have to synchronize the focus and exposure on every image. I solved part of the exposure issue to some extent in the Prague Gigapixel image (while it appears the Dresden folks did not in theirs) – you can see that the overall brightness of the image changes as you zoom in. I won’t reveal exactly what I did there 😉 But it’s clear that I did something, and it allows the details when zoomed in to have better contrast than they would otherwise.

The limitations

The main problem in creating these massive images is that we’re reaching the edge of certain optical properties of the camera and the lens. The camera can only make a single exposure, with a limited amount of dynamic range, which in practical terms is insufficient; the higher the resolution, the more insufficient it is.

Even more crucial than that is the focus and the depth of field (DOF).

It is not really noticeable in the Prague 18 Gigapixel image, shot with a 200mm lens, because everything is more than 100m away. In the 26 gigapixel dresden image, however, you can see that the horizon is not sharp – they have wisely chosen to focus the image not on infinity but rather on things that are much closer that you can actually see. With a 400mm lens, your DOF is really very small indeed. Even if you stop down to f/22 and use hyperfocal focusing, your DOF is still tiny. And you can forget about shooting more nearby things such as interiors because the DOF will be comparatively even smaller.

Focus Bracketing

The solution happens to exist already – it’s focus bracketing. It’s possible to shoot an identically composed photo which is focused differently, and to merge only the parts that are focused into a single image. This works great for single images. But what about stitching multiple images that are the result of focus bracketing?

Using this technique, it would be possible to use a very long lens to shoot hundreds of images for a gigapixel panorama, but to have a large DOF also. It would probably require 5 or 7 differently focused images at each position (and if exposure bracketing is used, you can multiply that by 3). This would limit the subject matter to things whose lighting won’t change over the course of many hours – but it would at least make a great deal more kinds of images possible at a huge resolution.

So, my questions:

1. Is it even possible to stitch multiple images that have been merged into a single focus-bracketed image?

2. Are there any cameras that exist which even allow automatic focus bracketing? I’m aware of this option in CHDK, but I don’t know how robust it is, and CHDK can only be used with “compact” cameras, not “pro” cameras.

3. Can an SLR tethered to a notebook or some other remote control perform focus bracketing?

Thanks to anyone for any light you can shed on these issues. In the meantime, maybe I’ll try some experiments myself and see what happens!